3186
J . Am. Chem. SOC.1989, I l l , 3186-3194
13000, Office of Basic Energy Sciences, Division of Chemical Sciences, and the National Institutes of Health, Biotechnology ~ Resource Program, ~ i ~of Research i ~ iR ~ ~ ~ we thank D ~ M.~ K,. id^^^^^ and s. p. cramer for their assistance with early experiments. Registry No. S, 7704-34-9;K2S04,7778-80-5;tetramethylene sulfone,
126-33-0; diphenyl sulfone, 127-63-9; dibenzothiophene sulfone, 101605-3; tetramethylene sulfoxide, 1600-44-8;2-methylthiophene, 554-14-3; benzothiophene, 132-65-0; methionine, ~ ~ 95- 15-8; ~ dibenzothiophene, ~ ~ ~ .6368-3; thianthrene, 92-85-3; dioctyl sulfide, 2690-08-6; tetramethylthiophene, 14503-51-6; benzyl phenyl sulfide, 83 1-91-4; 2-naphthalenethiol, 91-60-1; cysteine, 52-90-4; diphenyl disulfide, 882-33-7; thiohemianthraquinone, 68629-85-6.
Equilibrium Dynamics in the Thallium( 111)-Chloride System in Acidic Aqueous Solution Istvan Banyai' and Julius Glaser* Contribution from The Royal Institute of Technology (KTH), Department of Inorganic Chemistry, S-100 44 Stockholm, Sweden. Received August 9, 1988
Abstract: Kinetics of ligand exchange in the thallium(II1)-chloride system in aqueous three molar perchloric acid solution was studied by measuring 2osTlNMR line widths at 25 OC. The rate constants for the reactions
TICIm3-"'+ *TlCI,'-" are kol = 4.5 X lo4 (7.8 for the reaction
X
kO2= 6.4 x 105 M-1
(1.7
lo3 at 0 "C), k12= 5.2
X
& *TICIm3-"'+ TlCI,'-"
lo4 (1.5
lo4), kz3 = 2.7
X
X
IO7 (3.3
X
lo6), k3, < 3
X
lo7 M-'s-';
22T1CI2+
TI3++ TIC12+ s-1
X
lo5), and for the anation reactions TICI,'"
+ C1-
k'.c,.i)
T1C1,+12-"
k',,, < 1 X lo8, k'12= 3.3 X lo8 (2.4 X lo8), k'23 = 1.3 X lo9, and k'34 = 4.7 X lo8 M-l s-'. The first type of exchange dominates at low chloride to thallium ratios (CItot/TItot)= R 5 2, the second type is only observed when the species TI3+ and TlCl2+ are present, whereas the third one dominates at all higher R values. The activation parameters for the reaction represented by kol are AH* = 49 ( 1 2 ) kJ mol-l and AS* = +12 (k0.3) J mol-' K-I. A mechanism for this reaction is suggested to be a dissociatively activated interchange process. The activation parameters for the reaction represented by k'34 are AH* = 6.6 kJ mol-' and A S * = -107 J mol-l K-I. An associatively activated interchange mechanism for this reaction is proposed. The stability constant for the complex T1C152-was estimated to be K 5 = [TIC152-]/{[TlC14-][Cl-]}= 0.5 M-'.
The complexes formed in the thallium(III)-chloride system are among the strongest metal-ion-chloride complexes, and their equilibria have been extensively studied.2-8 There is information on their structures both in solid and in ~ o l u t i o n . ~ -However, ~~ there are only a few results on the kinetics of the complex formation and ligand exchange reactions of the thallium(II1) ion."+" ( I ) Permanent address: Department of Physical Chemistry, Lajos Kossuth University, H-4010 Debrecen, Hungary. (2) Lee, A. G. The Chemistry of Thallium; Elsevier: Amsterdam, 1971; pp 44-91, and references therein. (3) Ahrland, S.; Grenthe, I.; Johansson, L.;Noren, B. Acta Chem. Scand. 1963, 17, 1567. (4) Ahrland, S.;Johansson, L. Acta Chem. Scand. 1964, 18, 2125. (5) Woods, M . J. M.; Gallagher, P. K.; Hugus, Z . Z . ; King, E. L. Znorg. Chem. 1964, 3, 1313. (6) Kul'ba, F. Y . ;Mironov, V. E.; Mavrin, I. F. Zh. Fiz. Khim. 1965, 39, 2595. (7) Glaser, J.; Biedermann, G. Acta Chem. Scand. 1986, A40, 331. (8) Glaser, J.; Henriksson, U. J . Am. Chem. SOC.1981, 103, 6642. (9) Davies, E. D.; Long, D. A. J . Chem. SOC.A 1968, 2050. (10) Biedermann, G.; Spiro, T. G. Chem. Scr. 1971, I , 155. ( 1 I ) (a) Spiro, T.G. Inorg. Chem. 1965, 4 , 731. (b) Zbid. 1967, 6 , 569. (12) Glaser, J.; Johansson, G. Acta Chem. Scand. 1982, A36, 125. (13) Glaser, J. Acta Chem. Scand. 1982, A36, 451. (14) Kawai, Y . ;Takahashi, T.; Hayashi, K.; Imamura, T.;Nakayama, H.; Fujimoto, M. Bull. Chem. SOC.Jpn. 1972, 45, 1417. (15) Funada, R.; Imamura, T.;Fujimoto, M. Bull. Chem. SOC.Jpn. 1979, 52, 1535. (16) Lincoln, S. F.; Sandercock, A. C.; Stranks, D. R. Aust. J . Chem. 1975, 28,
1901.
(17) Henriksson, U.; Glaser, J. Acta Chem. Scand. 1985, A39, 355.
0002-7863/89/ 151 1-3186$01.50/0 , ,
I
The reason is probably that these reactions are rather fast not only on the traditional kinetic time scale but also on the N M R time scale of common N M R nuclei, e.g., ' H or 35Cl.16 Although there is a lot of data for the ligand and solvent exchange for the octahedral three-valent metal ions,I8 an investigation of dynamics in the thallium(III)-chloride system seems to be interesting. Structural studies showed that the geometry of thallium-halide complexes is different depending on the number of coordinated ligand^.'^-'^ In aqueous solution the aquated thaHium(II1) ion, the pentaaqua monochloro, and tetraaqua dichloro complexes are probably octahedral, the trichloro complex is tetrahedral or trigonal bipyramidal or both in equilibrium, but the fourth complex is clearly tetrahedral. These changes in the structure should influence the ligand exchange kinetics. In the case of AI3+, Ga3+, and In3+ the mechanism of solvent exchange shows a trend to turn from a dissociatively into associatively activated process when the ionic radius increases.19 Although the mechanism of solvent exchange is sometimes different from the mechanism of ligand exchange, in many cases similar trends have been observed. Kawai et al. studied the complex formation between Ga3+, In3+,and TI3+and semi-xylenol orange14 and later between TI3+and 4-(2-pyridylazo)resorcinol,15 by stopped-flow technique and found that the second-order rate (18) Wilkins, R. G. The Study of Kinetics and Mechanism of Reactions in Transition Metal Complexes; Allyn and Bacon: Boston, 1974. (19) van Eldik, R. Inorganic High Pressure Chemistry Kinetics and Mechanisms; Elsevier: Amsterdam, 1986.
0 1989 American Chemical Society
J . Am. Chem. Soc., Vol. 1 1 1 , No. 9, 1989 3187
Equilibrium Dynamics in the Tl(III)-Chloride System constants for the T1(OH)2+ species are of the same order of magnitude for both ligands. This observation was explained assuming that the rate-determining step is dissociation of a water molecule from the coordination sphere of TI(OH)2+. For TI3+, the rate constants were, unfortunately, not possible to identify. Lincoln et a1.I6 studied chloride exchange between the free chloride and GaCI4-, InC14(H20)2-, and probably respectively, by measuring j5CI N M R relaxation time in solutions containing very high chloride excess, [HCI] = 7 and 11 M. For the GaCI4- species associative mechanism was suggested while for InC14(H20)2-dissociative activation seemed to be probable. In the case of TICl6j- only a lower limit for the pseudo-first-order rate constant was estimated. Recently, Henriksson and Glaser" N M R signals for TI3+and T1C12+ in acidic observed separate 205Tl aqueous solution which shows that the chloride exchange between these two species is slow on a time scale defined by the difference between their chemical shifts. For higher complexes ( n 1 2) one signal was observed, but the line width was found to be controlled by chemical exchange. These observations allowed to determine lifetimes for the chloride and bromide exchange between the different Th(II1)-halide complexes. In the present paper we present kinetic results for the TI(111)-chloride system in aqueous 3 M HC10, solution based on 205TIN M R line width measurements. The total metal ion conM and the centration was varied between 5 X lo4 and 5 X ligand/metal ratio between 0 and 7.
Experimental Section Materials. About 1 M solution of TI(CI04)3was obtained by anodic oxidation of TIC104 solution.2h*bStock solutions containing 0.1 M total thallium were prepared at total chloride/total thallium ratio 0, 4, and 7. The total hydrogen ion concentration was 3 M, the ionic medium was I = 3 M (H+,(CIO,- + Cl-)]. The samples for measurements were prepared by mixing the different stock solutions and diluting with 3 M perchloric acid in order to obtain different total thallium concentrations. Analysis. TI(1) was determined by titration with 0.1 M solution of KBr03 using methyl orange as indicator. Total thallium content was obtained by reducing TI(II1) with SO2,boiling off the SO, excess, and titrating with 0.1 M KBrO3.,' Chloride content was determined by the Volhard method after reducing TI(II1) to TI(1) with SO,, removing excess SO2 by boiling, and cooling the solution. Prior to back-titration with KSCN, nitrobenzene was added in order to protect the precipitated AgC1.22 Acid concentration of TI(II1) solutions was determined by titration with NaOH after adding excess of solid NaCl to the analyzed solution. I n this way, the concentrations of acid, TI(1) and TI(II1) could be determined one after the other in the same sample. NMR Measurements. 205T1NMR spectra have been recorded at 230.8 MHz and at a probe temperature of 25 & 0.5 OC using a Bruker AM400 spectrometer. In order to obtain activation parameters some spectra have been recorded at different temperatures from 0 OC up to 60 OC. The NMR parameters were chosen so that approximately quantitative spectra could be obtained, typically: flip angle -25O (15 ~ s ) ,pulse repetition time = 0.1 s, sweep width -60000 Hz, number of scans = 5000-30000. The chemical shifts are reported in ppm toward higher frequency with respect to an aqueous solution of TICIOl extrapolated to infinite dilution. The accuracy of the measured line widths is about 10% and that of the chemical shifts about 0.1-1 ppm depending on the line width. Results and Calculations Determination of Chemical Shifts. The individual chemical shifts for the aqua and first complexes were measured directly from the spectra. For the chemical shift of the second complex a value was reported to be very close to that of the first one.8 Since fast exchange is expected between them," the accurate determination of their chemical shifts is critical from kinetic point of view because 1 ppm error in the chemical shift can mean more (20) (a) Biedermann, G . Arkiu Kemi 1953, 5,441. (b) Glaser, J. Thesis, The Royal Institute of Technology (KTH), Stockholm, Sweden, 1981. (c) Glaser, J. Acta Chem. Scand. 1979, ,433, 789. (d) Glaser, J. Acta Chem. Scand. 1980, ,434, 7 5 . (e) Glaser, J. Acta Chem. Scand. 1980, ,434, 141. (f) Glaser, J.; Johansson, G. Acta Chem. Scand. 1981, ,435, 639. (21) Noyes, A. A.; Hoard, J. L.; Pitzer, K. S.J . Am. Chem. SOC.1935, 57, 1231.
(22) Kolthoff, I. M.; Sandell, E. B. Texrbook of Quantitative Inorganic Analysis; Macmillan: New York, 1961; p 546.
1
a
b
I
I
I
I
I
I
I
2220
2200
2180
2180
2140
2120
2100
G(PPm) Figure 1. ,05TI NMR spectra at R = 1.12 and [Tilt,, = 0.02 M: (a) at 25 OC and (b) at 0 OC. The chemical shifts are reported in ppm toward higher frequency with respect to an aqueous solution of TIC104 extrap-
olated to infinite dilution. than 20% error in the N M R time scale and consequently high error in the rate constants. A t the total ligand/metal ratio R = 1.12 (throughout the text, R is used for the total chloride to the total thallium mole ratio) all the three above-mentioned species are present in measurable concentrations as can be calculated from the stability constant^.^ A t 25 "C only one *O5TIN M R signal is observed for 0.020 M solution (the intensity of aqua signal is very low and the peak is very broad because of chemical exchange), but a t 0 "C three signals appear (cf. Figure 1 ) . From this spectrum the chemical shift of the second complex was found to have 5.2 ppm lower value than that of the first one. The chemical shift difference between the first and the second complex was also measured a t higher R values at 0 OC and was found to be constant and equal to 5.2 ppm until R = 1.7, indicating that this value is not affected by exchange processes. On the other hand, it was also found that the chemical shift difference between the zeroth and first complexes is constant when the temperature is varied from 0 OC up to 60 OC. O n the basis of these facts we have accepted that the chemical shift difference between the second and the first complex is 5.2 ppm a t 25 O C , Le., the same value as found a t 0 OC. For the third and fourth complexes the individual chemical shifts were calculated with the following equation: where 'hOM is the experimental chemical shift for the experimental point i, iPn is the molar fraction, and 6, is the chemical shift of
3188 J. Am. Chem. Soc., Vol. 111, No. 9, 1989
Banyai and Glaser
Table I. 205TINMR Chemical Shifts" and Nonexchange Line Widths for TlCI,.'-" Complexes in 3 M HCIO,, at 25 OC Au1,20 (this 6' (ppm) work)b (Hz) 6 (ref 8)c n (in TICI,'-") 0 2093 ( f l ) 25 (11) 2086 ( f l ) 1 48 (5) 2198 (14) 2204 ( f l ) 2 2199 ( f l ) 25 2201 (2) 3 2435 (4) -45 2412 (5) 4 2648 (5) -25 2645 (2) 5 2022 ( f 2 0 ) d 2022 (*20)d "The chemical shifts are reported in ppm toward higher frequency with respect to an aqueous solution of TIC10, extrapolated to infinite dilution. Numbers in parentheses with f are maximum errors of measurements, otherwise standard deviations obtained from leastsquares calculations. bContains magnetic field inhomogeneity. c I n (3 M HClO, + 1 M NaClO,), 27 OC. dDetermined by measuring wide line 205TlNMR for powder samples.
a
b
AV; ( Hz )
1000
-
the nth complex. The p , values were calculated using the stability constant (p,) values published by Woods et al. for the same ionic medium.' 6, values were determined by least-squares method using the experimental chemical shifts in the range 1.8 < R < 7. In this region the chemical exchange can be considered fast enough for eq 1 to be valid. The chemical shift values are given in Table I and compared to the values determined earlier.8 From the tendency of the change of the chemical shift with ligand/metal ratio it has been found that the fifth complex also forms in the system when R > 6. Since there is no stability data for this species in three molar perchloric acid we estimated it from our measurements using the chemical shift value determined by Glaser and Henriksson for the solid NazTlCl5.12H2O8and applying eq 1 for the calculation. The best fitting was obtained for log Osequal to 18.02, that is for log K 5 = -0.3. Determination of Empirical Rate Equations. 0 C R C 1.0. For chloride/metal ratios 0 < R < 1.0 the exchange between TI3+and T U 2 +is slow on the actual time scale. Hence, measured line widths for each of the signals can be written as23
500
500
0
0
0
0.02
0.04
0
0.02
0.04
[T13'l,n, (MI
Figure 2. Dependence of line width of TIS+(a) and TIC12+(b) on [TI],,, at R = 0.24 (O), R = 0.5 (A), R = 0.7 (0).T = 25 OC.
0.5
0
1 .o
R
(2)
Figure 3. Dependence of line widths (0of TI3+and 0 of T1Cl2+) and molar fractions (dashed line) of complexes on total ligand/metal ratio at [Tl],,, = 0.01 M. Full line represents line widths calculated by using the rate constants in Table 11. T = 25 OC.
where A V , /is~ the measured line width and 1/T2eXP is a sum of three terms
different coordinated species are constant independent of the total thallium concentration. Introducing these conditions into eq 4
A v I / ~= 1/(rTzeXP)
1 / TzexP= 1/ Tz
+ 1 / Tinh+ 1
/Tex
(3)
containing the transverse relaxation time, the inhomogeneity of the magnetic field, and the chemical exchange contribution. In eq 3 only l / depends ~ ~ on~the concentration of different species present in solution. Thus, the measured line widths of the signals of TI3+ and T1C12+contain pseudo-first-order rate constants for all reactions in which these species take part. At first approximation these reactions can be of zero or of first order for other species present, that is n
= Av,/:(i)
n
+ ci + j=O~ c ' ~ ~ [ C +I - j=O ]x c i j p j
(4)
where 4vll?(i) = l / T z + l / T p h (nonexchange line width for the ith complex), p j is the molar fraction of the j t h complex, c denotes constants which are proportional to the rate constants, index i means that the equation belongs to the signal of the ith complex, i n d e x j indicates the complex species to which the rate is proportional, and c ' ~denotes rate constants of the reactions for which the rates are proportional to the ith complex and the free ligand concentration yielding the j t h complex. Equation 4 can be applied to obtain the constants c by least-squares analysis in the case of experiments when the total thallium concentration is constant but R is varied. Equation 4 can also be modified for least-squares analysis of experiments when the ligand/metal ratio is constant but the total thallium concentration is varied. Since the complexes formed in this system are very strong, at constant R the free chloride concentration and the relative populations of (23) Sandstrom, J. Dynamic N M R Spectroscopy; Academic Press: Lon-
don, 1982; pp 1-29.
aul/2
=
IR
+ SRIT1ltot
(5)
where ZR = 4vl/2(i) + c, + CY=~C'~~[CI-], SR= zy,oc,,p,, and the index R indicates the total ligand/metal ratio. The dependence of line widths on the total thallium concentration at different R values has been found to be linear (cf. Figure 2). The intercepts for line widths of the aqua complex do not depend on R which means that no free chloride contributed reaction occurs in the exchange process of the aqua complex, that is c b l = 0. The mean value of the intercepts is 25 f 11 which is very close to the measured line width for the aqua complex in absence of chloride ions (22 f 3 Hz). Consequently, it can be assumed that the nonexchange line width for the aqua complex, Le., 4 ~ , / ~ ~ =( 025) Hz. For the T1C12+ signal the intercepts increase slightly with increasing R, but the change is not significant enough in order to calculate c',~.The intercept for TIC12+at R = 0.24 is accepted as 4vli20(1) = 48 & 5 Hz for the monochloro complex. From Figure 2 it can also be seen that the slopes strongly depend on the total ligand/metal ratio indicating the presence of exchange reactions with contribution of complexed species. In order to obtain the constants c,, the line width versus R data have been considered a t constant [TI],,, (cf. Figure 3). From the line widths of the TI3+signal, using the nonexchange line width determined above from the total concentration dependences, the values of parameters coI = 314 f 20 and cO2= 2043 f 330 have been found by the least-squares method on the basis of eq 4. For the line widths of the T1C12+signal two sets of parameters have given a fitting with approximately the same goodness. In the first set col = 258 & 8, cI1= 68 f 16, and c12= 917 f 230 are supposed to describe the system; in the second set col = 254 i 9, c I I = 49 f 18, but instead of c I 2(which corresponds to reaction 23, vide
J. Am. Chem. Soc., Vol. 111, No. 9, 1989 3189
Equilibrium Dynamics in the Tl(III)-Chloride System Table 11. Rate Constants for Different Exchange Reactions at 25 OC used in the model (M-I s-l) rate constants (M-' s-I) 4.5 X IO4 (4.9 f 0.41 X IO4." (4.1 f 0.3) X k,, kO2 (6.4 f i . i j x 105 6.4 x 105
".
k,, kIz k23
(9.4 f 2.7) X I O 3 (5.1 f 0.4) X 104,c(5.3 f 0.3) X (2.7 f 0.7) X IO'
5.2 X IO4
k',2
(3.3 f 0.4) X IO8,' (3.1 f 0.5) X IO8" (1.3 f 0 . 2 ) X I O 9 (4.7 f 1.3) X IO8
3.2 X IO8 1.3 x 109 4.7 x 108
k'23
k'34
2.7
x 107 '
From the peak of the aqua complex. bFrom the peak of the first complex. 'From R = 1.3. "From R = 1.5. infra) we obtain c ' , ~= (2.4 f 0.5) X lo8 (corresponds to reaction 24). Since the constants indexed 12 belong to an exchange between T1C12+and TIC12+ and are only determined by some experimental points a t R > 0.7, it is difficult to choose between the two sets. For the exchange between the aqua and first complexes the following reactions can be proposed on the basis of the calculations above:
+
kol e * ~ 1 3 ++ TICP+
~ 1 3 + *TICP+
(6)
This is a reaction without any net chemical change but with change in the magnetic environment of the thallium nuclei. This work deals with a chemical system in equilibrium. Hence, the concentrations are not changing in time, and all terms of the type d[T1CIn3-"]/dt are equal to zero. However, in order to calculate the connections between the dynamics in the investigated solutions and the N M R spectra we will use usual rate equations = k[A][B] ...+...etc. but with a different of the type d[TlCI:-"]/dt meaning. The left side of such an equation will represent the rate of the microscopic change of the concentrations of the different species. Thus, the rate corresponding to reaction 6 is d[TI3+]/ d t = 2kOl[TI3+][TICI2+]
(7)
because one aqua complex forms in the forward and one in the back reaction. The rate constant kol can be calculated from col by using eq 2 and 4 as kol = cola/(2[T1],). From cozthe following exchange path can be proposed
TI3++ TICI2+
2T1Cl2+
(8)
which leads to net chemical change. The rate equation is -d[TI3+] / d t = ko2[T13'] [TIC12+]
(9)
and the value of kO2= C ~ ~ T / [ T I ] ~The , , . rate constants are given in Table 11. From line widths of TICl2+the constant coI belongs to reaction 6, and from Table I1 it can be seen that the calculated rate constant agrees with the previously calculated value of k o l . Considering the reaction 8, cI1can be proportional to the rate constant of the reverse path. The rate equation is d[TI3+]/dt = -j/,d[T1ClZ+]/dt = kll[T1C1z+]2
(10)
-_ 1
c
'
,\:
2
3
d[T1CI2+]/ d t = 2kll [TICI2+I2
(11)
and k , , = clla/(2[Tl]lo,) where cI1= 60 as an average value of the two sets. Certainly, there should be a connection between kO2 and k , I determined by equilibrium constants k02/kll
= [T1C12+]2/[T13+][T1C12+] = 01'/@2
(12)
From our experimental rate constants ko2/kll = 66, while from = 53 which is a good the potentiometric equilibrium data5 @12/p2 agreement taking into account that the standard deviations of the determined rate constants are 15% for kO2and -25% for k l l . Finally, the slopes of line width for the zeroth and first chloride
-
-----.o
4
R
Figure 4. Dependence of line widths (0)and molar fractions (dashed line) of TICl,,-" complexes on total ligand/metal ratio at [TI],,, = 0.02 M. Full line represents line widths calculated by using the rate constants in Table 11. T = 25 OC.
complexes, respectively, vs [TI],,, can be calculated from the rate constants a t different ratios with eq 5. At R = 0.24, S p ~ =~ ~ l ~ 7200 (experimental = 6462 f 250) and SRiCalcd = 23 100 (22060 f 184); a t R = 0.5, SpCalCd = 15 200 (16 450 f 300) and SpCalcd = 17800 (16630 f 176); a t R = 0.7, Spcalcd = 24800 (23410 f 852) andSRP"" = 12900 (12407 f 420). Thus, the agreement with the measured values seems to be excellent. R > 1. When the total ligand/metal ratio is higher than 1 (at 25 "C) always one signal appears in the spectra indicating occurrence of fast exchange on the time scale determined by the chemical shifts of the different thallium complexes. The dependence of line widths and the molar fractions of the complexes on R is shown in Figure 4. The bandshape of an N M R spectrum can be described by the following equation adopting the formalism introduced by Reeves and S h a ~ ~ ~ V = (constant).C1.P (13) where P is the column vector of molar fractions and C is a matrix C = R2 + W . C ~ * W (14) where R2 matrix is the sum of diagonal matrix of A v l 1 2 ( i ) and K matrix which contains the pseudo-first-order rate constants, cz = R2-I, w is a diagonal matrix of frequency variable wj = x - Qj, where Qj are the chemical shifts for the different species, and x is the frequency variable. For creation of K rate matrix there is an excellent example in the literature.26 In our case it can be seen from Figure 4 that at certain ratios the exchange system can be considered as three sites exchange. For example, at ratio higher than 2.5 but lower than 6 the exchange process can be described by the following scheme: TICI2+
For this scheme the rate equation is d[TIClj] /df = K[TICIj]
that is
,
,
(15)
where the left hand side is the column vector of exchange rate, [TICI,.] is the column vector of concentrations, and the form of K rate matrix is the following:
with the usual connection between the rate constants pikij = p,k-!) From eq 13 and 1 4 it follows that the line width in this case IS (24) Piette, L. H.; Anderson, W. A. J . Chem. Phys. 1959, 30, 899. (25) Reeves, L. W.; Shaw, K. N. Can. J . Chem. 1970, 48, 3641. (26) Chan, S. 0.;Reeves, L. W. J . Am. Chem. SOC.1973, 95, 670
3190 J . A m . Chem. SOC.,Vol. 111, No. 9, 1989
Banyai and Glaser a
3i
b
3
c is Y
0 ‘ 2
2-
X”
1
0
0.06
0.03
W) Figure 5. Dependence of the pseudo-first-order rate constant k340bad (cf. eq 17) on free chloride concentration at [TI],,, = 0.005 M (n),[Tl],,, = 0.01 M (O),[TI],,, = 0.02 M (A), [TI],,,= 0.03 M (0). T = 25 O C .
0
[CI’I
determined by three pseudo-first-order rate constants, three chemical shifts, and three values of Avli;. Actually, in our system no direct exchange of the type TlCl2+ + *T1CI4-
& T1C12+ + *T1C14-
has been found to occur. The chemical shifts values are known. The AvljZovalues for the different complexes can be estimated in the following way. At 0.05 and 0.03 M total thallium concentration at R = 7 it was found that the line width is 25 H z independently of the composition; that is, the exchange is very fast. Since at these circumstances 90% of the thallium in the solution is present in the form of T1CI4-, we can assume that for the species TlCl4- Au1/2(4) 25 Hz. In very concentrated thallium solutions a t R = 2 where the relative population of the second complex is about 80% the line width was found to be 25 Hz. Similarly, a t R = 3 the third complex constitutes 70% of the total thallium, and the line width of the signal a t high [TI],,, values is 40 Hz. From these experiments we accepted Avl12(2) 25 Hz for TlCl2+and A~,~20(3) 40 Hz for T U 3 . Since the experimental line widths are usually higher than 400 Hz, the error introduced by these estimations is within the uncertainty of the measured line widths. At R values higher than 4 it was assumed that only the exchange between the third and the fourth complex affected the line widths. Consequently, it can be considered as two sites exchange. All experimental spectra have been successfully fitted as Lorentzian curve. For this case Piette and Anderson deduced an approximate formula to obtain pseudo-first-order rate constants directly from the experimental line
-
-
-
k340bsd=
47V%p42(6v)2/[Av1p
- p3Avi/z0(3) - P4Avijz0(4)] (16)
where 6v is the difference between the individual chemical shifts of third and fourth complex in Hz. Assuming first-order de(calculated by eq 16) on the free chloride pendence of l~~~~~~~ concentration (k’34)and on the concentration of the TICI; species (k34)the following equation can be deduced to express the change of k34 as a function of the free chloride and [T13+],,,
+ k34K4[T13+]tot[C1-](l+ K4[C1-])-’ (17) where K4 = p4/p3and [TI3+],,, = [TICI,] + [T1C14-]. However, kjqobsd= k;,[Cl-]
the plot of k340bsdversus free chloride shows a linear dependence with zero intercept independent of [TI3+], (cf. Figure 5). Hence, there is no significant contribution of exchange with direct collision of TICI, and T1CI4-, and from the slope of the straight line in Figure 5 , k’34 can be calculated. Thus, the dominant exchange path between TICI, and T U 4 is TIC13
+ CI-
k 54
k’&
TIC14-
(18)
1 0.04
0 0
0.08
0.04
0.08
[T13’1,,, (M)
Figure 6. Dependence of kl2Obsd(a, cf. eq 22) and k230bad (b, cf. eq 19) on [TI],,, at R = 1.3 (a,A), R = 1.5 (a,O), and R = 2.52 (b). T = 25 OC.
Supposing maximum 10% contribution to the line width of the reaction path with direct collision of T1Cl3 and T1CI4- one can find an upper limit for k3, C 3 X l o 7 M-’ SKI, In order to obtain the rate constants for the exchange between TlCl2+ and T1CI3 the dependence of the line width on the total thallium concentration has been considered at R = 2.52 using eq 13-15 in the following way. In the K rate matrix, k’34[C1-] has and kZ3Obsd values have been varied been substituted for k340bsd, until the experimentally measured line width was obtained. The k230bsd values found show linear dependence on the total thallium concentration as can be seen in Figure 6. The following equation is suggested to describe this function: k230bsd
= 2k 23P3 [TI] tot + k ’23 [c1-1
(19)
that is, the following exchange paths can be said to be dominant: TICI2+
+ *T1CI35TICI, + *T1CI2+
(20)
and T1CI2+
+ CI-
kb k’-z
TU3
Since the complex distribution as well as the free ligand concentration in the system change less than 8% when the total thallium concentration changes from 0.012 to 0.05 M, the free chloride concentration and the molar fraction of the third complex can be considered approximately constant. Accepting this assumption k’23 has been calculated from the intercept and k23 from the slope of the straight line in Figure 6 as (1.3 f 0.2) X lo9 M-’ s-l and (2.7 f 0.7) X lo7 M-’ s-I, respectively. By using these values of the rate constants the dependence of the line width on the total ligand/metal ratio at 0.02 M total thallium concentration has been calculated by eq 5 for 2.5 IR 5 4. The agreement between the calculated and experimental line widths is rather good, as can be seen in Figure 4. For 1 < R < 2.5 four site exchange, involving complexes from the aqua up to the third complex, has been considered when eq 5 was used in order to find the rate constants for the exchange between T1C12+and TIC12+. At first the dependence of line width on [TI],,, has been described at R = 1.32 and 1.52 in the same way as at R = 2.52 by using the rate constants which have already been determined. At both ratios linear dependence was found for k120bsdvs [TI],,, as it is shown in Figure 6, that is
k120bsd= 2klflpz[T1]tO,+ k112[CI-]
(22)
Hence, the following reaction paths are proposed to be dominant:
+ *T1CIZ+2T1C12++ * T U 2 +
(23)
T1C12+ + C1- -!% T U 2 +
(24)
TU2+ and
‘kl-12
J . A m . Chem. Soc., Vol. 111, No. 9, 1989 3191
Equilibrium Dynamics in the TI(III)-Chloride System
a
b
AVi (Hz)
Pi
6004 200 1.0
100
0.5
0
,
0.04
0.02
0 0
0.02
0.04
[TI"],,, (M)
R
Figure 7. Dependence of the line width of TI3+ (O), TICI2+ (a),and TICI2+(A)signals and molar fractions of complexes (dashed line) on the total ligand/metal ratio at [TI],,, = 0.02 M at 0 OC. The full line represents line widths calculated by using the rate constants in Table 111.
Figure 8. Dependence of the line widths of TIC12+(a) and TICI2+(b) signals on [TI],,, at R = 1.3 at 0 O C .
\
\ \
Table 111. Rate Constants for Different Exchange Reactions at 0 values (M-I s-l) values (M-' s-I) kol ( 7 . 8 i 2.3) X IO3 k2, (3.3 f 0.8) X lo6 kO2 ( 1 . 7 0.2) x 105 k'12 (2.4 f 1.2) X IO8 k,,
OC
(1.5 f 0.6) X IO4
The rate constants are k 1 2= (5.3 f 0.3) X lo4 and (5.1 f 0.4) lo4 M-l S-I and k'12 = (3.1 f 0.5) X lo8 and (3.3 f 0.4) X lo8 M-I s-l for R = 1.32 and R = 1.52, respectively. By using average values of the rate constants, dependence of line width on R at [TI],,, = 0.02 M has been described as it is shown in Figure 4. Summarizing, Figures 3, 4, and 5 show that our model reflects the principal character of the system with the rate constants collected in Table 11. Measurements at 0 "C. From Figure 1 it can be seen that the exchange reactions between TI3+, T U 2 + ,and TICI2+ at 0 "C and at [TI],,, = 0.02 M are so slow that the signals of all species present can be observed separately (up to the ratio 1.7). It allows us to calculate the rate constants directly from the line width data by eq 2-4. In Figure 7, dependence of the line width and the population of the different complexes is shown as a function of R. In order to calculate the distribution curves a t 0 O C the enthalpies of formation for the different species have been used.s At R = 1.3 the total concentration dependence of the line width has also been measured (cf. Figure 8). Analyzing the dependence of the line width of the aqua complex on R by least-squares method, the values of A U , / ~ ~ (= O )50 f 18 Hz, kol and ko2 have been determined (cf. Table 111). Since slow exchange takes place in the measured range of R the linewidth of T1CI2+is expectably determined by five parameters, i.e., Avli20(l ) , kol, k l l , k,,, and k'12using the same letters as at 25 OC. Independent parameter fitting for all five parameters have been unsuccessful because of the small number and the high relative error of the experimental data. However, there is a possibility to determine k , , independently from the total concentration dependence a t R = 1.3. From eq 5 the slopes of the straight lines (cf. Figure 8) are
41
"\I
X
and sR2
=
c02P02
+ c12P1 + c23P3
(26)
From eq 25 using coIand cI1= 2c0fi2/PI2values determined above, cI2and consequently k , , can be calculated. The numeric value of k 1 2is 1.4 X lo4 M-I s-l. Assuming that a t R = 1.3 the term ~ 2 !,negligible, 3 ~ from eq 26 k , , has been found to be 1.6 X lo4 M- s in good agreement with the previous value. Now, using kol, k , , , and k I 2the values Av,,?(l) = 27 f 6 H z and k',2 have
3.0
3.5 1 / T ~ 1 0 3OC')
Figure 9. Arrhenius plot of the rate constant dependence on the temperature: (0)for kol;(0)for k;4. been determined by least-squares analysis on the basis of eq 4. The rate constants are given in Table 111. There is a possibility to check the and k'12 values because from eq 5 the intercept of total concentration dependence of the line width of T1C12+should be equal to I,] = ( A v , / ~ O1)( k'12 [CI-]/a). The calculated value of IR] is 97 Hz; the experimentally found value is 84 i 17 Hz. The line width of T1CI2+ versus R data are expectably determined by five parameters (Au1,?(2) + k'-12),k02, k12, k23, and k'23,of which only the first and the last two are unknown. The least-squares parameter fitting has shown that the dependence of the line width on R (for R 5 1.7) can be approximated by assumption of k23 rather than k'23 giving the numerical values (Av1/?(2) + kLI2/a) = 89 f 15 H z and k23 = (3.3 f 0.8) X lo6 M-' s-l. Plotting the line width of TlC12+vs [TI],,, a t R = 1.3, a straight line is obtained (cf. eq 5 and Figure 8b), and the intercept can be deduced as KR2 = ( A u , / ~ O ( + ~ )kLI2/a). The numeric value of this parameter has been found to be 117 f 19 Hz, compared to the value obtained from R dependence (89 f 15 Hz); it seems to be a reasonable agreement. It is also possible to calculate the value of kCI2 as kLI2 = k'lfiI/P2 = 440 f 220 s-I considering the fi values without error. It can be seen that dividing this value by a one can obtain 140 H z which is larger than the sum (Av1120(2) + kLI2/r); that is negative value would be obtained for A V ~ , ~ O ( The ~ ) . probable reason for this irresonable result is the high uncertainty of k'12 determined from R dependence. The Temperature Dependence. At the ratios 0.4 and 6, respectively, we have only one dominant exchange path. Therefore, studying the temperature dependence there is a possibility to calculate activation parameters and by this way to obtain more knowledge about the mechanisms. In Figure 9, the temperature dependence of k,, is shown in an Arrhenius plot. The activation
+
3192 J . A m . Chem. SOC.,Vol. 111, No. 9, 1989 w,,(%)
Banyai and Glaser
[
Scheme I TI(H20)e3'
I
0
\
2
3 I d
R
Figure 10. Percentage contribution of different parallel exchange paths (w,,, numbered by the indices of rate constants) to the total exchange rate w,/ = ~,,(2k,[TICI~] [TICI,] + k',,[TICI,] [Cl-1). T = 25 "C.
enthalpy is 49 f 2 kJ mol-', and the activation entropy is +I2 f 0.3 J mol-' K-I. At ratio 6, it can be seen from Figure 9 that the line width becomes relaxation-controlled when the temperature is increased above 40 OC. Approximate activation parameters for the reaction represented by are AH* N 6.6 kJ mol-' and AS* N -107 J mol-'. Furthermore, there is a possibility to obtain approximate activation parameters for the remaining reactions a t 25 OC and 0 "C. These values are certainly very rough but may be of some (though limited) value for discussion of the mechanism(s) (see below). For the reaction paths characterized by koz,AH* 32 kJ mol-' and AS* -27 J K-l; kl,, AH' 33 kJ mol-' and ASs -45 J K-I; k'12,AH* 7.6 kJ mol-' and A S * -56 J K-I; k23, AH* 51.2 kJ mol-' and AS* 69 J K-I.
-
- -
T1CI(H20)52'
I
34'/
1
+
-
-
-
-
Discussion By the above interpretation of the experimental results seven reaction paths are suggested to be dominant in the TI3+-CI-system depending on the concentrations of the components (cf. Figure IO). The reaction paths denoted k,, (cf. reactions 6, 20, and 23) are the type of chemical exchange reactions in which no net chemical change occurs. This reaction type can be called selfexchange using the nomenclature introduced for a similar type of electron-transfer reactions.19 Reaction 8 also represents an exchange between complex species, but the microscopic process includes net chemical change. Usually, this type of ligand exchange reactions cannot be observed because of their low rate constants compared to the anation reactions such as reactions 18, 21, and 24 (Table 11). In the present case the high stability of the coordinated species results in a very small concentration of the free ligand at lower ligand/metal ratios and allows the selfexchange reactions to become dominant. Moreover, anation reaction of TI3+ has not been observed. In Figure 10, relative participation of the different parallel exchange paths in the total exchange rate is plotted as a function of the total ligand/metal ratio. It can be seen that at 0.7 IR I1.5 six parallel paths exist with comparable rate. Certainly, the contribution of these exchange paths to the values o f t h e line widths is different because beside the rate constants the chemical shift differences between the complexes which are present affect the line width governed by fast-exchange regime.23 Model calculations have shown that the line width values a t 0.7 IR I1.5 are determined by exchange of all species except TIC1,- and are very sensitive for the difference in the chemical shifts between TICIZ+and TICI,', which are the dominant species at these ratios. For example, when the value of the chemical shift of TICI2+ is lowered by 1 ppm, k , , and k'12can be found to be higher by 50% and 20%, respectively. Since the chemical shift of TICI2+ has been extrapolated from 0 O C , 0.3 ppm uncertainty is a reasonable assumption so that the accuracy of rate constants for the exchange between T1C12+ and T1C12+ is lower than the calculated standard deviations show. This fact is taken into account in the further discussion.
A or
Ia
Second-order rate equations have been found for the self-exchange reactions. Consequently, both species should take part in the transition state(s). Formation of transition-state species can be imagined either by the decreasing or, more probably, by the increasing of the coordination number of the central thallium atom, because the ionic radius of TI3+N 95 pm and for the smaller In3+ ion (80 pm) transition state for water exchange reaction is obtained by the increasing of the coordination number (according to ref 19). On the other hand, the size of the coming ligand can also determine the mechanism. The following scheme shows these two possibilities for the reaction between the aqua and the monochloro complexes. In Scheme I and throughout the discussion thallium(II1) chloride complexes are assumed to coordinate water molecules according to the results obtained earlier by means of X-ray diffraction on s ~ l u t i o n , ' ~on . ' ~crystals,20bfIR Raman,9Jl and N M R data,8 Le., T1(H20):+, T1CI(H20)s2+,TIC12(H20),+,T1Cl3(H20) or T1C13(H20),, and T1C14-. The coordinated water molecules are always written in the formulas when they play an important role in the actual considerations, otherwise they are omitted for the sake of simplicity. The left hand side reactions in Scheme I are analogous to a dissociatively activated ligand exchange mechanism, while the right hand side to an associatively activated one; but there are important differences. Since our rate equations are first order for both species which take part in the exchange reaction, the limiting dissociative mechanism (D) can be closed out. The interchange mechanisms, according to the Eigen-Wilkins mechanism,2' proceed in two steps: the fast formation of an outer-sphere complex, described by an equilibrium constant K,, followed by the rate-determining outer-sphere to inner-sphere interchange step, characterized by the rate constant krds. In the case of I d mechanism of simple ligand exchange the dissociation of the leaving ligand is thought as the rate-determining step (koldin Scheme I). But in the present type of self-exchange the rearrangement and the decomposition of the intermediate (kolr in Scheme I) can also be imagined as the rate-determining step. In spite of this special property of self-exchange we shall use the symbol of dissociative interchange Id, for this mechanism. In the case of the associative part of Scheme I, KO,rather represents the formation of activated complex ( K * ) if the mechanism is limiting associative (A) and the rate-determining step is the formation of the seven-coordinated intermediate (kolf). Langford and Gray28call this mechanism limiting A if the kinetic test shows the presence of an intermediate of increased coordination number. When no intermediate species can be found, as in our case (unless the intermediates are very short-lived, which is not impossible), the mechanism is declared an interchange process. Consequently, the associative interchange, I,, (rather than A) seems to be the reasonable alternative taking into account kolfor kOldeC as krdsafter the outer-sphere complex formation. However, there are some arguments which rather support the I d mechanism than the I,: (i) Positive values of activation entropy (27) Wilkins, R. G.; Eigen, M. Ado. Chem. Ser. 1965, 49, 5 5 . (28) Langford, C. H.; Gray, H. B. Ligand Substitution Processes; W. A. Benjamin Inc.; New York, 1965.
J . Am. Chem. SOC.,Vol. 111, No. 9, 1989 3193
Equilibrium Dynamics in the Tl(III)-Chloride System found for the exchange reaction between the T1(H20)63+and T U 2 +and roughly estimated for the self-exchange between TICI2+ and T1C13 show that the transition state is less ordered than the initial one. (Rough estimation leads to a negative value of the activation entropy for the self-exchange between the T1C12+ and TICI2+ as well as between TI3+and TICI2+, but because of the uncertainty of this parameter it cannot be stated that the two latter exchange reactions are of the type of associative interchange in spite of the fact that the two former reactions are not. Here it should be emphasized that the activation parameters include an unknown contribution from the formation of the outer-sphere complex. In general, this fact lowers the predictive potential of the activation entropies. In this particular case, however, the opposite situation probably occurs: the formation of the outersphere complex is a kind of associative process and in spite of the complication arising from the rearrangement of the solvent molecules, the formation probably gives a negative contribution to the calculated activation entropies. Hence, if the calculated AS* for the reaction is positive, the entropy contribution from the rate-determining step is certainly positive thus indicating a dissociative type of reaction mechanism.) (ii) The entering "ligand", which in this case is another complex species (e.g., T1C1(H20)52+), is so large that its coordination as a seventh ligand seems less probable. In the light of above considerations the rate equation for the self-exchange reactions can be written as rate = krdrKo,[TIC1~-"] [T1C1,+12-"] (27) that is the rate constants determined experimentally are the multiplication of the outer-sphere equilibrium constants and the rate constants of the rate-determining step (krds). There is a possibility to estimate KO, values for the different direct exchange paths using the Bjerrum-Fuoss equation,29which assumes only electrostatic interactions to be present
KO,= 4rNa3/3000 exp (-2)
(28)
where N is Avogadro's number, a is the encounter complex separation being 5 A (one layer), and 2 is the coulombic energy divided by k T . Following numerical values can be calculated: K,(O,l) = 7.5 X M and krds(O,l) = 6 X los s-l; K0,(0,2) = 4.7 X M and krds(0,2) = 1.4 X lo8 s-'; K,(1,2) = 1.2 X M and k*( 1,2) = 4 X 1O6 s-l; K,(2,3) = 3 X 10-I M and k*(2,3) = 9 X IO7 s-I. In the case of Id mechanism krdsshould be within one order of magnitude of the rate constant of water exchange and should be independent of (or only slightly dependent on) the nature of the entering ligand. Unfortunately, no experimental data are available in the literature for water exchange on TI3+, but there is an estimation of kold> 3 X lo9 s-l based on chloride substitution reactions of Fe3+ complexes.30 Furthermore, Lincoln et a1.16 found by 35C1NMR measurements that the rate constant of CI- exchange reaction between T1C163-and free chloride is larger than 1.3 X lo6 and 1.6 X lo6 SKI,in 7 and 11 M HCI, respectively. It can be accepted as a lower limit for the TI-CI bond breaking and thus for the rate constant for the rearrangement of the intermediate. The calculated rate constant for the rate-determining step of the self-exchange between TI3+and TIC12+, krds(O,l)= 6 X lo8 s-l, is lower than the lower limit estimated for the water exchange rate constant for the aqua complex.30 Since both the lower limit of water exchange rate constant30 and our calculations for koldcontain simplifications, neither the difference nor the agreement is significant. The same mechanism (Id) is thought for the reaction between TI3+and TlC12+(eq 8) in spite of the (roughly estimated) negative value of the activation entropy. The numerical values of k*(O,l) and krds(0,2)are of the same order of magnitude. Since the chloride moves from one complex to another in a self-exchange reaction, the most probable picture of the transition state is a binuclear species containing chloride ligand between the two thallium atoms. It means that in the self-exchange reactions 6
and 8 the water should leave the aqua complex part of the ion pairs before the formation of the intermediates. That is, if the rate-determining step in these two reactions is the cleavage of the water-thallium bond, k*(O, 1) and k*(0,2) should have the same or a t least similar values. Thus, for the self-exchange reactions in which TI3+takes part, koldseems to be the rate-determining step. In the self-exchange reaction between TIC1" and TlCl2+ or between T1C12+and TICI3a water molecule can leave both sides of the ion pairs yielding binuclear complexes with chloride bridge ligand. W e assume higher rate constant values for the water exchange rate for these complexes than for that of the aqua complex because the water-thallium bond is longer20band hence presumably more labile in the former complexes. It means that the rearrangement of the transition binuclear complex can become the rate-determining step instead of the dissociation of water. The extremely low value for the krds(1,2) may be attributed to a very symmetric intermediate containing an axial C1--T13+